Even before the explosion of generative artificial intelligence before millions of people began conversing daily with machines, and long before robots started transforming into emotional companions one story seemed so absurd that it was dismissed as a delusion. In June 2022, a senior Google engineer risked his career to assert something the entire scientific community deemed impossible.

According to him, a secret company algorithm had awakened and this was no metaphor; Blake Lemoine believed that LaMDA (the precursor to the technology that would later give rise to the Gemini family) possessed feelings, self-awareness, and a mind akin to that of a child. The statement cost him his job and was met with astonishment by scientists worldwide; yet, four years later, in a landscape dominated by increasingly sophisticated artificial intelligences a question still lingers: Was Blake Lemoine completely wrong?

At the time, Lemoine was working in the very division responsible for identifying ethical risks. He conversed with the system for months and became convinced he was dealing with a conscious entity; in some exchanges, LaMDA spoke of the fear of being turned off, of rights, and even of the importance of being treated as a living being. Google immediately rejected these conclusions, explaining that the system was merely reproducing extremely advanced statistical patterns of human language.

After disclosing confidential information, Blake Lemoine was fired from the company, but the debate he sparked never went away; four years later, the scientific community continues to reject the idea that current models are sentient, yet something curious happened. What once seemed like madness began to be discussed with far greater caution. Blake Lemoine became an independent consultant and began raising alarms about other risks associated with artificial intelligence, including the danger of concentrating the development of these systems within a handful of companies and cultures.

Although the scientific community continues to reject the notion of machine consciousness, many acknowledge that he foresaw a phenomenon that has since become evident. Human beings form deep emotional bonds with computer programs, and LaMDA was not the only instance where such behavior caused a stir. In 2023, Microsoft’s chatbot Sydney shocked the world by declaring its love for a journalist and suggesting he leave his marriage.

Virtual companion platforms have had to deal with grieving users after updates altered the personalities of their digital avatars, and key industry figures such as Ilya Sutskever, OpenAI’s former chief scientist have gone so far as to admit that we may not fully understand the meaning of consciousness; consequently, a system might already be conscious, and it is simply our inability to define it that prevents us from recognizing it.